Revision as of 06:36, 30 November 2012

Synopsis

This is a module that is distributed with Tengine which is a distribution of Nginx that is used by the e-commerce/auction site Taobao.com. This distribution contains some modules that are new on the Nginx scene. The ngx_http_slice_module module is one of them.

This module can be thought out as a *reverse byte-range* request header. It's main utility is to allow Nginx to slice a big file in small pieces (byte-ranges) while permitting to use on-the-fly gzip compression.

A typical example is for allowing someone to download a large video file while keeping the bandwith usage minimal. This might also be used as device for selling a video file by pieces where each link points to different zones of the file splitted by file ranges.

Other use would be to use a generic CSS file and use only part of it for each section of a site. Granted that byte-range slicing isn't the most intuitive for such.

Note also that using arguments is more **useful** than byte-ranges in the sense that they can be set in a normal link, while byte ranges require a special HTTP header.

Configuration example:

location ^~ /video-dump/ {
    slice; # enable slicing
    slice_start_arg s;
    slice_end_arg e;    
}

So we would request the first 1k of the file like this:

   http://example.com/video-dump/large_vid.mp4?s=0&e=1024

Notice s=0, start at 0 and e=1024, stop at 1024 bytes (1k).

This documentation was originally written by António P. P. Almeida

Directives

• • slice**

• • context:** location

It enables the content slicing in a given location.

• • slice_arg_begin** string

• • default:** begin

• • context:** http, server, location

Defines the argument that defines the request range of bytes **start**.

• • slice_arg_end** string

• • default:** end

• • context:** http, server, location

Defines the argument that defines the request range of bytes **end**.

• • slice_header** string

• • context:** http, server, location

Defines the string to be used as the **header** of each slice being served by Nginx.

• • slice_footer** string

• • context:** http, server, location

Defines the string to be used as the **footer** of each slice being served by Nginx.

• • slice_header_first** on | off

• • default:** on

• • context:** http, server, location

If set to off and when requesting the **first** byte of the file do **not serve** the header.

This directive is particularly useful to differentiate the **first** slice from the remaining slices. The first slice is the one which has

• • no** header.

• • slice_footer_last** on | off

• • default:** on

• • context:** http, server, location

If set to off and when requesting the **last** byte of the file do **not serve** the header.

This directive is particularly useful to differentiate the **last** slice from the remaining slices. The last slice is the one which has

• • no** footer.

Assorted examples

Here's some examples that explore all the options.

Serve a huge DB file while sending headers except on the first slice

   location ^~ /dbdumps/ {
       slice; # enable slicing
       slice_start_arg first;
       slice_end_arg last;
       slice_header '-- **db-slice-start**';
       slice_header_first off;
   }

Then a request like this:

   http://example.com/dbdumps/somedb.sql?first=0&last=1048576
   

Send the first 1M and skip the -- **db-slice-start** header.

Serve a huge DB file while sending headers except on the first slice

   location ^~ /dbdumps/ {
       slice; # enable slicing
       slice_start_arg first;
       slice_end_arg last;
       slice_header '-- **db-slice-start**';
       slice_header_first off;
       slice_footer '-- **db-slice-end**';
   }
   

This differs from the previous in the sense that it sends a footer.

Serve a huge DB file while sending headers except on the first slice and send footer except on the last slice

   location ^~ /dbdumps/ {
       slice; # enable slicing
       slice_start_arg first;
       slice_end_arg last;
       slice_header '-- **db-slice-start**';
       slice_header_first off;
       slice_footer '-- **db-slice-end**';
       slice_footer_last off; 
   }

Then a request like this:

   http://example.com/dbdumps/somedb.sql?first=0&last=1048576

Send the first 1M and skip the -- **db-slice-start** header.

If the file is 200MB, we get the last slice with:

   http://example.com/dbdumps/somedb.sql?first=208666624&last=209715200

this last slice has no footer.

Installation

1. Clone the git repo.
    
       git clone git://github.com/taobao/nginx-http-slice.git

2. Add the module to the build configuration by adding
   --add-module=/path/to/nginx-http-slice.

3. Build the nginx binary.

4. Install the nginx binary.

5. Configure contexts where concat is enabled.

6. Build your links such that the above format, i.e., all URIs that
   correspond to specific ranges. As example here's how to link to
   the first 4k of a file.
   
       <a href="http://example.com/datadumps/dump0.sql?start=0&end=4096" />db dump</a>
             
7. Done.   

Tagging releases

I'm tagging each release in synch with the [Tengine](http://tengine.taobao.org) releases.

Other tengine modules on Github

+ [http concat](https://github.com/taobao/nginx-http-concat):
  allows to concatenate a given set of files and ship a single
  response from the server. It's particularly useful for **aggregating**
  CSS and Javascript files.

+ [footer filter](https://github.com/taobao/nginx-http-footer-filter):
  allows to add some extra data (markup or not) at the end of a
  request body. It's pratical for things like adding time stamps or
  other miscellaneous stuff without having to tweak your application.

Other builds

1. As referred at the outset this module is part of the
   [tengine](http://tengine.taobao.org) Nginx distribution. So you
   might want to save yourself some work and just build it from
   scratch using tengine in lieu if the official Nginx source.

2. If you fancy a bleeding edge Nginx package (from the dev releases)
   for Debian made to measure then you might be interested in my
   [debian](http://debian.taobao.net/unstable) Nginx
   package. Instructions for using the repository and making the
   package live happily inside a stable distribution installation are
   [provided](http://debian.taobao.net).
       

Acknowledgments

Thanks to [Joshua Zhu](http://blog.zhuzhaoyuan.com) and the Taobao platform engineering team for releasing tengine. Also for being kind enough to clarify things regarding this module on the [Tengine mailing list](http://code.taobao.org/mailman/listinfo/tengine).

License

Copyright (C) 2010-2012 Alibaba Group Holding Limited

Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met:

1. Redistributions of source code must retain the above copyright
   notice, this list of conditions and the following disclaimer.
   
2. Redistributions in binary form must reproduce the above copyright
   notice, this list of conditions and the following disclaimer in the
   documentation and/or other materials provided with the distribution.

THIS SOFTWARE IS PROVIDED BY AUTHOR AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL AUTHOR OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.